How to see and control which apps use generative AI in Windows 11

If you use Windows 11 and have started using artificial intelligence tools, you've probably wondered at some point which apps are using those resources exactly. generative AI models that are integrated into the systemMicrosoft is putting AI practically everywhere: in File Explorer, in Copilot, in third-party apps… and it's important to understand what's happening "behind the scenes" so you don't lose control of your data or your privacy.

Furthermore, with the arrival of new privacy options in Windows 11, it is possible to see Which applications have recently accessed the system's generative AI modelsas well as better managing which AI tools are used in a personal or corporate environment. This is complemented by advanced solutions such as Microsoft Purview (DSPM for AI) and Defender for Cloud Apps, designed primarily for companies that want to monitor and limit the use of generative AI applications within their organization. We're going to learn all about it right now. How to see which apps have recently used generative AI models in Windows 11.

AI Actions in Windows 11 File Explorer

Microsoft is testing some new options in Windows 11 called AI actions integrated into File Explorerdesigned so you can work with images and documents, even if you manage them in a private AI gallery, without having to open them in external programs.

These actions allow you to perform the following with a right-click: quick editing tasks on image files, such as retouching photographs, removing unwanted objects, or blurring the background to focus attention on the main subject.

Within these functions there is also a specific action for Perform reverse image searches using the Microsoft search engineso that you can find similar or related content on the Internet to the photo you have selected.

According to the Windows team, with these AI actions in Explorer, the user can interact more advancedly with your files from the context menu itself.so you can edit images or summarize documents without breaking your workflow.

The underlying idea is that you can stay focused on your tasks while You delegate the heaviest editing or analysis tasks to AI.avoiding opening several different applications for very specific things.

For now, these new features are not available to everyone, since Only users enrolled in the Windows Insider program can test them., Microsoft's early testing channel.

If you are part of that program, you can activate these features by right-clicking on a compatible file and selecting the option “Artificial intelligence actions” in the Explorer context menu.

Currently, these actions are being deployed in the Canary Channel with the Windows 11 Build 27938, a very early, test-oriented versionTherefore, it is normal for there to be changes and adjustments over time.

New privacy section: Which apps use generative AI in Windows 11

With that same build, Microsoft has incorporated a New section within Settings > Privacy and security dedicated exclusively to text-to-image generation and the use of generative AI models by applications.

This section shows it clearly. Which third-party applications have recently accessed Windows' generative AI models?This is especially useful if you're concerned about security or want to know which programs are using AI resources without your full awareness, including those accessible from browsers like Sidekick.

Thanks to this panel, users can better control which apps have permission to use these AI capabilities, adjusting access in a similar way to how it is done with the camera, microphone or other sensitive permissions.

With these types of controls, Microsoft reinforces its commitment to integrate artificial intelligence natively into the operating systembut at the same time providing tools so that the user does not lose sight of privacy and data management.

Advanced management of the use of generative AI applications in companies

Beyond home use, in corporate environments it is essential that security teams can detect, monitor and control which AI applications are being usedwhether they are from Microsoft or belong to other providers.

Microsoft has designed a strategy to defense in depth around Microsoft 365 Copilot and other proprietary AI solutionswith multiple layers of security to protect data, identities, and regulatory compliance.

The big question that arises is what happens to artificial intelligence applications that are not from Microsoftespecially those based on generative models that employees can access from the browser.

To address this scenario, Microsoft offers tools such as Data Security Posture Management (DSPM) for AI within Microsoft Purview and Microsoft Defender for Cloud Apps (part of the Microsoft Defender family) that allow security departments to more strictly manage the use of AI apps.

With these solutions, the goal is to give organizations the ability to to use generative AI applications in a safer and more controlled waythus reducing the risk of exposure of sensitive information or non-compliance with regulations.

Why monitoring AI applications is key

Monitoring and managing generative AI applications has become essential for minimize data leaks, maintain compliance, and implement appropriate governance about how these technologies are used, for example when using local models.

In practice, this means that the organization must be able to to detect which AI services are being used, what type of information is being sent, and what risks are involvedespecially when it comes to confidential or regulated content.

Microsoft proposes using DSPM for AI and Defender for Cloud Apps together for detect, monitor and, if necessary, block or limit generative AI applications, relying on cloud application policies and catalogs.

Using DSPM for AI (Microsoft Purview) to discover and manage AI applications

DSPM for AI, integrated into Microsoft Purview, offers security and compliance teams visibility into the activity involving the use of generative artificial intelligence dentro de la organización.

With this tool it is possible protect data that are included in requests to AI services and to exercise greater control over how that data is handled and shared, something critical when users upload internal documents to Chatbots or similar services. OneDrive with artificial intelligence It is an example of AI integration with user data within the Microsoft ecosystem.

The first recommendation is create or activate AI-specific Purview policiesDSPM for artificial intelligence includes pre-configured policies that can be enabled with very little effort.

These "one-click" directives allow you to define clear rules about what types of data can or cannot be involved in interactions with generative AI applicationsthus reducing the possibility of accidental exposure.

Once the policies are implemented, it can be seen Generative AI-related activity in the Activity Explorer and audit logs, which provides a detailed and traceable history.

These records include, for example, User interactions with generative AI sites and services accessible from the browser, allowing us to understand what tools employees are experimenting with.

Events are also recorded in which Data loss prevention (DLP) rules are triggered during the use of AI appsThis indicates attempts to share sensitive data with external services.

The system also reflects when they have detected types of confidential information in those user interactions, making it easier for security personnel to identify risky behaviors.

As a complement, it is highly recommended Configure DLP policies specific to the Microsoft Edge browserso you can protect navigation from uncontrolled AI services while also taking advantage of Copilot's AI mode in Edge.

Through these policies, it is even possible Block access to unmanaged AI applications from unprotected browsersthus forcing traffic to pass through monitored channels.

Using Microsoft Defender for Cloud Apps with generative AI applications

Microsoft Defender for Cloud Apps provides an additional layer of control by allowing detect, monitor or block generative AI applications used in the organization, relying on a catalog of cloud applications with risk scores.

From the Microsoft Defender portal you can access a catalog of categorized cloud applications, including the “generative AI” category, which groups all apps of this type detected in the environment.

By filtering by that category, security teams obtain a list of generative AI applications along with their security and compliance risk scoresThis helps to prioritize which services should be analyzed in depth.

These scores are calculated by combining different factors, making them useful for decide which apps are worth monitoring more closely or even blocking if they do not meet the organization's requirements.

Create a policy to monitor generative AI applications

Within Defender for Cloud Apps, you can define specific policies for monitor the use of new generative AI applications detected in the organization, as part of a continuous control model.

First, it's important to make sure the prerequisites are met and review the documentation on cloud application control through custom policiesbecause the configuration is flexible.

When creating a new policy, one usually starts from an empty template, selecting “No template” as the policy type to be able to adjust all parameters manually.

A name can be assigned to the policy that makes its purpose clear, for example “New applications of generative AI”, and set a medium severity level (such as level 2) to calibrate the alerts.

The directive description should explain that An alert will be generated each time a new generative AI application is detected and used., thus facilitating its identification by the security team.

In the conditions section, it is common to state that the The application must belong to the “generative AI” category.so that policy focuses solely on this type of service.

Finally, the policy can be configured to apply to all continuous cloud application discovery reportsensuring that detection covers all monitored traffic.

Create a policy to block certain AI applications

In addition to monitoring, Defender for Cloud Apps allows block specific AI applications that the organization considers unauthorized, applying governance measures to its use.

Before that, it's advisable to review the documentation on cloud application control and governance policy creation, since this type of policy can have a direct impact on users.

The process usually begins in the section of Cloud apps > Microsoft Defender Portal Cloud Discovery, where the applications detected in the organization are listed.

Within that view, you can apply the filter of “Generative AI” category to show only applications of this typethus facilitating their analysis and selection.

In the results list, select the AI ​​application you want to restrict, and in its row, the options menu will appear. assign it the label of an “unauthorized” or “unsanctioned” app, officially marking it as blocked at the governance level.

Next, in the navigation panel, you can access the section of cloud application governance to manage associated policies, including those that will apply to apps labeled as unauthorized.

From the policies tab, a new custom policy is created by selecting again “No template” as a configuration basis, so that tailored criteria and actions are defined.

Politics can be called, for example, “Unauthorized AI applications” and be described as a rule intended to block generative AI applications labeled as unauthorized.

In the conditions section, you can specify that The application category is generative AI and the label is Unsanctioned, limiting the scope exactly to what you want to block.

Once this is configured, the policy applies to all ongoing app discovery reportsensuring that traffic to those apps is identified and blocked according to established rules.

Basic control of recently installed applications in Windows 11 and Windows 10

Although the focus is on AI, it can also be useful to know What applications have been recently installed on your Windows 11 PC?For example, to identify possible AI-related programs that you don't remember installing.

In Windows 11, you can quickly open settings by typing “Apps and features” in the taskbar search bar and clicking on the corresponding result to access the list of apps.

Within that section it is possible Change the sorting criteria to “Installation Date”, which makes the most recent applications appear at the top of the list.

If you want the search to be more precise, you can use the option to “Filter by” and select “All drives” to cover all disks, or choose a specific drive if you know where the program is installed.

The applications will then be displayed ordered by the date they were last installed in the systemalong with relevant information such as the version, which is useful for checking new installations.

In each record you can expand the icon of More options to access actions such as uninstalling the application directly, if you notice something that doesn't convince you.

You can also use the box of Search for applications within that same screen to locate a program by name or keyword.This speeds up management if you have many apps installed.

In Windows 10 the procedure is very similar: simply search for “Apps and features” from the search bar and open the corresponding settings panel.

From there, you again have the option to sort by “Installation Date” and filter by unitAnd when you select an application, you can see its version or delete it if you deem it necessary.

Similarly, you have a field for Search the list by typing the name or a term related to the applicationshowing only the matching results.

AI-generated descriptions in model-based applications

In the realm of enterprise applications, Microsoft is also leveraging AI to Generate automatic app descriptions based on models, with the aim of helping users better understand what each application does.

Complex applications can be confusing for end users, so AI analyzes the app's content and structure to Create a clear description that explains its main functionality..

The header for these apps and the app switcher have been updated with a more modern style designed to integrate these AI-generated descriptionsso that when interacting with the application name, this explanatory text is displayed.

When the app creator does not manually add a description, the system may Generate it automatically using integrated AI models, displaying the result both in the header and in other parts of the interface.

In the application designer, the owner can View the generated description, accept it as is, or modify it.adjusting it if it detects that context is missing or there are nuances that need clarifying.

If the explanation includes AI-generated content and the creator chooses not to accept it yet, the app may display a notice or disclaimer indicating the origin of that description, which adds transparency to the process.

Quick ways to find apps in Windows

Beyond the settings panels, Windows offers simple shortcuts for search for installed applications or specific programs when you need them, which is very useful if your menu is full.

The most direct way is to Use the search button on the taskbar and type the name of the application or program., letting the system suggest the shortcut without having to navigate through menus.

Another equally quick option is Press the Windows key on your keyboard and start typing the app name directlybecause the Start menu behaves like a built-in search engine.

With these gestures, you can locate in seconds recent programs, AI tools, or any application you want to openeven if you don't remember exactly where it's anchored.

With all these pieces, it's clear that Microsoft is strongly pushing the integration of AI into Windows 11 and its ecosystem, but at the same time offering increasingly more options for See which applications have recently used generative AI models, regulate their access, and better manage security risks.both in personal devices and in corporate environments where control and traceability are fundamental.